• Keine Ergebnisse gefunden

Dynamics and chemistry of vortex remnants in late Arctic spring 1997 and 2000: Simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS)

N/A
N/A
Protected

Academic year: 2022

Aktie "Dynamics and chemistry of vortex remnants in late Arctic spring 1997 and 2000: Simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS)"

Copied!
11
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Atmos. Chem. Phys., 3, 839–849, 2003

www.atmos-chem-phys.org/acp/3/839/

Atmospheric

Chemistry and Physics

Dynamics and chemistry of vortex remnants in late Arctic spring 1997 and 2000: Simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS)

P. Konopka1, J.-U. Grooß1, S. Bausch1, R. M ¨uller1, D. S. McKenna2, O. Morgenstern3, and Y. Orsolini4

1Institute for Stratospheric Chemistry (ICG-I), 52425 J¨ulich, Germany

2National Center for Atmospheric Research, Boulder, CO, USA

3Max-Planck-Institut f¨ur Meteorologie, Hamburg, Germany

4Norwegian Institute for Air Research (NILU), Kjeller, Norway

Received: 30 December 2002 – Published in Atmos. Chem. Phys. Discuss.: 25 February 2003 Revised: 4 June 2003 – Accepted: 16 June 2003 – Published: 23 June 2003

Abstract. High-resolution simulations of the chemical com- position of the Arctic stratosphere during late spring 1997 and 2000 were performed with the Chemical Lagrangian Model of the Stratosphere (CLaMS). The simulations were performed for the entire northern hemisphere on two isen- tropic levels 450 K (≈18 km) and 585 K (≈24 km).

The spatial distribution and the lifetime of the vortex rem- nants formed after the vortex breakup in May 1997 display different behavior above and below 20 km. Above 20 km, vortex remnants propagate southward (up to 40N) and are

“frozen in” in the summer circulation without significant mixing. Below 20 km the southward propagation of the rem- nants is bounded by the subtropical jet. Their lifetime is shorter by a factor of 2 than that above 20 km, owing to sig- nificant stirring below this altitude. The behavior of vortex remnants formed in March 2000 is similar but, due to an ear- lier vortex breakup, dominated during the first 6 weeks after the vortex breakup by westerly winds, even above 20 km.

Vortex remnants formed in May 1997 are characterized by large mixing ratios of HCl indicating negligible, halogen- induced ozone loss. In contrast, mid-latitude ozone loss in late boreal spring 2000 is dominated, until mid-April, by halogen-induced ozone destruction within the vortex rem- nants, and subsequent transport of the ozone-depleted polar air masses (dilution) into the mid-latitudes. By varying the intensity of mixing in CLaMS, the impact of mixing on the formation of ClONO2 and ozone depletion is investigated.

We find that the photochemical decomposition of HNO3and not mixing with NOx-rich mid-latitude air is the main source of NOxwithin the vortex remnants in March and April 2000.

Ozone depletion in the remnants is driven by ClOxphotolyt- ically formed from ClONO2. At the end of May 1997, the halogen-induced ozone deficit at 450 K poleward of 30N Correspondence to: P. Konopka

(p.konopka@fz-juelich.de)

amounts to≈12% with≈10% in the polar vortex and≈2%

in well-isolated vortex remnants after the vortex breakup.

1 Introduction

Long-term ground-based measurements over Europe show that the total column of ozone began to decline in the 1970s.

The decrease was greatest in the winter/spring period, with the ozone decline trend in the late 1990s being around 3–

6%/decade. The main contribution is located in the 12–20 km altitude range (WMO, 1998).

One of the widely discussed mechanisms contributing to this trend is the ozone depletion in the Arctic vortex and its impact on the mid-latitudes. During winter and spring, chlorine in the polar vortex is activated on the surface of po- lar stratospheric clouds and causes severe ozone destruction (WMO, 1998). Knudsen and Grooß (2000) estimated from calculations done for 1995 and 1997 that approximately 40%

of the observed TOMS total ozone trends result from trans- port of the ozone-depleted vortex air into the mid-latitudes.

By applying this method north of 63 for the period 1992–

2000, Andersen and Knudsen (2002) deduced that the major part (75%) of the observed Arctic ozone decrease in March and most of the variability may be explained by winter/spring ozone depletion in the polar vortex. On the other hand, using 3-D CTM studies Chipperfield and Jones (1999) showed that during the 1990s the dynamic variations dominated the inter- annual variability of ozone north of 63with little evidence of a trend towards more wintertime chemical depletion.

The first studies considering the details of the transport of the vortex air into the mid-latitudes were based on LIMS data (LIMS – Limb Infrared Monitor of the Stratosphere) and GCM simulations. Hess (1991) found that long-lived anomalies of tracers were still observed two months after c

European Geosciences Union 2003

(2)

840 Konopka et al.: Dynamics and chemistry of vortex remnants the breakup of the polar vortex in spring 1979. Using the

PDF technique for 3-D simulations of N2O, Orsolini (2001) identified some long-lived westward-propagating tracer pat- terns in the 1998 summer polar stratosphere above 20 km that resulted from the slow advection of coarsely-mixed vortex remnants. Based on 3-D CTM simulations with a parameter- ized ozone chemistry, Piani et al. (2002) showed that by the end of June, 2000, above 420 K, much of the ozone-depleted air was transported from the polar regions to the subtropics, whereas below 420 K, these air masses remained polarward of≈55N. They suggested that below 420 K the subtropical jet provides an effective transport barrier while stirring after the breakup of the polar vortex is important at upper levels.

Despite these achievements, questions concerning the life- time, the spatial distribution, and the year-to-year variability of such remnants are still open. Also their impact on the total ozone column, in particular over the densely populated re- gions in northern mid-latitudes, needs to be quantified. Fur- thermore, only few studies exist that discuss the effect of stir- ring and mixing on the ozone chemistry in the slowly diluting vortex remnants (Marchand et al., 2003).

Inaccurate representation of mixing in photochemical transport models may influence the predictions of strato- spheric ozone depletion (Edouard et al., 1996; Searle et al., 1998a,b). Satellite observations (Riese et al., 1999), in situ measurements (Tuck, 1989), and dynamical model studies based on such experiments (e.g. Plumb et al., 1994; Waugh et al., 1997; Orsolini et al., 1997; Balluch and Haynes, 1997) demonstrated the existence of filamentary structures on a broad range of spatial scales in stratospheric chemical tracer fields. Chemical transport models that do not resolve fil- amentary structures and do not represent their dissipation times realistically will not simulate non-linear chemical reac- tions accurately. Using a photochemical box model to study the impact of mixing on the deactivation of stratospheric chlorine, Tan et al. (1998) showed that both box models with- out mixing and the currently employed grid-based numerical models can, in certain circumstances, significantly over- and underestimate the observed ozone loss rates. Furthermore, based on idealized isentropic simulations, they concluded that a spatial resolution better than 40 km is necessary to ob- tain a correct description of ozone loss that is not sensitive to the numerical diffusivity of the model.

To study the impact of transport on the spatial distribution, the lifetime, and the ozone chemistry in the vortex remnants in spring/summer 1997 and 2000, we use the high-resolution, isentropic (2-D) version of the Chemical Lagrangian Model of the Stratosphere (CLaMS) (McKenna et al., 2002,a). We chose these two periods because of their completely different characteristics with respect to the lifetime of the Arctic polar vortex: Whereas the vortex in 1997 was extremely long-lived with the final breakup around mid-May (Coy et al., 1997), the vortex decay in mid-March 2000 is more typical for the final warmings observed in the last twenty years (Manney and Sabutis, 2000).

The Lagrangian view of transport allows mixing to be de- scribed in its own way. Mixing in CLaMS is induced by an adaptive regridding of the (isentropic) air parcels (APs) with mean distance to the next neighbors given byr0(model reso- lution). The regridding procedure is applied after each advec- tion time step1t (6–24 hours) that is conducted in terms of isentropic (2-D) trajectories. The mixing intensity is driven by the horizontal deformations in the flow measured by the finite time Lyapunov exponentλ. Significant mixing occurs only in flow regions where λ exceeds a prescribed critical valueλc. In contrast, in the Eulerian approach the numerical diffusion is present always and everywhere due to the high frequency of interpolations of fluid elements on the fixed spa- tial grid (Courant criterion), the contribution of the CLaMS regridding procedure to the transport can be contolled. It can be continuously reduced (e.g. by increasing the critical Lya- punov exponentλc) until tracers are solely transported along trajectories without any mass exchange between the APs. We call this (reversible) part of transport pure advection. The re- gridding procedure is controlled by the critical Lyapunov ex- ponentλc, the model resolutionr0, and the length of the ad- vection time step1t(or the grid adaptation frequency 1/1t).

These parameters define the (irreversible) part of transport, i.e. mixing.

The paper is organized as follows. In Sect. 2 we describe the model configuration and justify the isentropic approx- imation by comparing the CLaMS CH4 distributions with HALOE observations. Section 3 discusses the spatial dis- tribution and the lifetime of the vortex remnants observed in spring and summer 1997 and 2000. In Sect. 4, the ozone chemistry occurring in these remnants is considered. The impact of mixing on the chlorine deactivation and the ozone loss is discussed in Sect. 5. Finally, conclusions are drawn in Sect. 6.

2 Simulations with CLaMS

Isentropic CLaMS studies are carried out for two periods:

10.04–31.07 1997 and 10.2-01.06 2000, at the two isentropic levels 450 and 585 K. The meteorological fields are taken from ECMWF data. As a reference case, we employ a (quasi- uniform) distribution of APs over the northern hemisphere with the distance between the neighboring APsr0 =65 km and r0 = 200 km north- and southward of 30N, respec- tively. The critical Lyapunov exponentλc and the time step 1t are set to 1.2 day−1 and 24 hours, respectively. In the following, we call this configuration of mixing parameters adjusted or optimal mixing. The optimal mixing leads to the best agreement between the CLaMS simulations and the in situ aircraft measurements of tracers observed at the edge of the northern polar vortex during the SOLVE/THESEO-2000 campaign (Konopka et al., 2003). The combinations of the mixing parameters can be quantified by the so-called effec- tive diffusivity (perpendicular to the wind direction) given as

Atmos. Chem. Phys., 3, 839–849, 2003 www.atmos-chem-phys.org/acp/3/839/

(3)

Konopka et al.: Dynamics and chemistry of vortex remnants 841 D≈r02/(41t )exp(−2λc1t )(McKenna et al. (2002) use the

notationDc for the effective diffusivityD discussed here).

For the optimal mixingDamounts to 1.1 103m2s−1north- ward of 30N.

In addition, to study the influence of mixing on chem- istry, we consider distributions of APs with spatial resolu- tions r0 = 100, 200 and 400 km northward of 30N. The strongest impact on mixing can be achieved by varying the spatial resolution due to the quadratic dependence of the ef- fective diffusivity onr0. Thus, the corresponding effective diffusivities are 2.6 103, 1.1 105 and 4.2 105m2s−1. The CLaMS transport scheme resolves horizontal structures up to the order ofr0exp(−λc1t ), i.e. forr0 =65 km the smallest resolved scales are approximately 20 km.

The initial distribution of all chemical species on 10.04.1997 is derived from MLS and HALOE observations using trajectory mapping, tracer/tracer and tracer/equivalent latitude correlations. In addition, ER-2, POAM and TRIPLE observations are taken into account for the initialization at 10.02.2000. The remaining species are initialized from a 2- D model climatology. The denitrification in 1997 is derived from the MLS data, whereas a one-month temperature his- tory is used to parameterize the sedimentation of NAT and ice particles and the subsequent effect of HNO3and H2O re- moval (Grooß et al., 2002; Konopka et al., 2003).

In Fig. 1 the isentropic CLaMS distributions of CH4at 450 and 585 K about 12 days after the vortex breakup in 1997 are shown. For comparison HALOE observations are over- laid (circles). Vortex remnants at 450 K are more strongly bounded around the pole and are mixed more intensively than the remnants at 585 K.

Before we quantify these properties more precisely, some remarks on the validity of the isentropic calculations are nec- essary. The isentropic approximation is motivated by weak diabatic descent of the vortex air at both levels between mid- February and the end of May. Diabatic descent rates derived from the radiation module (Zhong and Haigh, 1995) based on the Morcrette scheme (Morcrette, 1991) show that during the period under consideration, the vortex air masses do not sig- nificantly change their potential temperatures in theθ-range between 400 and 600 K (1θ <30 K).

Aditionally, we verify the quality of the CLaMS transport by comparing the simulated distributions of CH4 with the HALOE observations. Figure 2 shows the results of such comparison for two CLaMS configurations: without mix- ing (i.e. transport only in terms of forward trajectories) and with the optimal mixing. The correlation coefficientρ be- tween the observed and simulated mixing ratios (see legend in Fig. 2) show that the CLaMS simulation with optimal mix- ing reproduces fairly well the CH4 distributions observed by HALOE in spring and summer 1997 northward of 30N. Fur- thermore, the CLaMS mixing scheme smoothes out some of the very low CH4 mixing ratios present in the pure advec- tion transport that are not observed by HALOE (in partic-

Konopka et al.: Dynamics and chemistry of vortex remnants 3

In addition, to study the influence of mixing on chem- istry, we consider distributions of APs with spatial resolu- tions

r0 =

100, 200 and 400 km northward of 30

0

N. The strongest impact on mixing can be achieved by varying the spatial resolution due to the quadratic dependence of the ef- fective diffusivity on

r0

. Thus, the corresponding effective diffusivities are 2.6 10

3

, 1.1 10

5

and 4.2 10

5

m

2

s

1

. The CLaMS transport scheme resolves horizontal structures up to the order of

r0exp(−λc∆t), i.e. forr0 =65 km the smallest

resolved scales are approximately 20 km.

The initial distribution of all chemical species at 10.04.1997 is derived from MLS and HALOE observations using trajectory mapping, tracer/tracer and tracer/equivalent latitude correlations. In addition, ER-2, POAM and TRIPLE observations are taken into account for the initialization at 10.02.2000. The remaining species are initialized from a 2D model climatology. The denitrification in 1997 is derived from the MLS data, whereas a one month temperature his- tory is used to parameterize the sedimentation of NAT and ice particles and the subsequent effect of HNO

3

and H

2

O removal (Grooß et al., 2002; Konopka et al., 2003).

In Fig. 1 the isentropic CLaMS distributions of CH

4

at 450 and 585 K about 12 days after the vortex breakup in 1997 are shown. For comparison HALOE observations are overlayed (circles). Vortex remnants at 450 K are stronger bounded around the pole and are mixed more intensively than the remnants at 585 K.

Before we quantify these properties more precisely, some remarks on the validity of the isentropic calculations are nec- essary. The isentropic approximation is motivated by weak diabatic descent of the vortex air at both levels between mid- February and the end of Mai. Diabatic descent rates derived from the radiation module (Zhong and Haigh, 1995) based on the Morcrette scheme (Morcrette, 1991) show that during the considered period, the vortex air masses does not sig- nificantly change their potential temperatures in the

θ-range

between 400 and 600 K (∆θ <

30

K).

Aditionally, we verify the quality of the CLaMS trans- port by comparing the simulated distributions of CH

4

with the HALOE observations. Fig 2 shows the results of such comparison for two CLaMS configurations: without mix- ing (i.e. transport only in terms of forward trajectories) and with the optimal mixing. The correlation coefficient

ρ

be- tween the observed and simulated mixing ratios (see legend in Fig 2) show that the CLaMS simulation with optimal mix- ing reproduces fairly well the CH

4

-distributions observed by HALOE in spring and summer 1997 northward of 30

0

N. Fur- thermore, the CLaMS mixing scheme smoothes out some of the very low CH

4

mixing ratios present in the pure advec- tion transport that are not observed by HALOE (in partic- ular at 585 K). Consequently, the correlation coefficient

ρ

between the HALOE observations and CLaMS simulations increases from 0.5 to 0.74 and from 0.38 to 0.64 for

θ=

450 and 585 K, respectively. The deviations still present at 585 K are mainly caused by the errors of the simulated absolute po-

Fig. 1. CH4 simulated with CLaMS at θ = 585 K (top) and 450 K (bottom) at 22.05.1997, i.e. about 12 days after the vor- tex breakup. The circles denote the HALOE observations (tangent points) mapped to the same synoptic time.

sition of the vortex remnants. This indicates that the quality of the ECMWF winds is better at 450 than at 585 K.

Thus, using 3d-trajectory calculations and by comparing CLaMS CH

4

-distributions with the HALOE observations, we conclude that isentropic transport dominated the propa- gation of the vortex air into the mid-latitudes several weeks before and after breakup of the vortex in 1997 and 2000.

3 Spatial distribution and lifetime of vortex remnants in 1997 and 2000

In 1997, the polar vortex was unusually long lived (Coy et al., 1997). Even at the end of April, the vortex was very symmet- ric around the pole and stable, and it did not break up until mid May. On the other hand, the breakup of the vortex in 2000 was around mid March, i.e. during a time period typical

www.atmos-chem-phys.org/0000/0001/ Atmos. Chem. Phys., 0000, 0001–12, 2003

Fig. 1. CH4 simulated with CLaMS at θ = 585 K (top) and 450 K (bottom) at 22.05.1997, i.e. about 12 days after the vor- tex breakup. The circles denote the HALOE observations (tangent points) mapped to the same synoptic time.

ular at 585 K). Consequently, the correlation coefficient ρ between the HALOE observations and CLaMS simulations increases from 0.5 to 0.74 and from 0.38 to 0.64 forθ =450 and 585 K, respectively. The deviations still present at 585 K are mainly caused by the errors of the simulated absolute po- sition of the vortex remnants. This indicates that the quality of the ECMWF winds is better at 450 than at 585 K.

Thus, using 3-D trajectory calculations and by comparing CLaMS CH4 distributions with the HALOE observations, we conclude that isentropic transport dominated the propa- gation of the vortex air into the mid-latitudes several weeks before and after breakup of the vortex in 1997 and 2000.

www.atmos-chem-phys.org/acp/3/839/ Atmos. Chem. Phys., 3, 839–849, 2003

(4)

842 Konopka et al.: Dynamics and chemistry of vortex remnants

4 Konopka et al.: Dynamics and chemistry of vortex remnants

0.5 1.0 1.5

CH4 (CLaMS) [ppmv]

Pure advection

585 K, ρ=0.38 450 K, ρ=0.50

585 K, ρ=0.38 450 K, ρ=0.50

0.5 1.0 1.5

CH4 (HALOE) [ppmv]

0.5 1.0 1.5

CH4 (CLaMS) [ppmv]

Advection + Mixing

585 K, ρ=0.61 450 K, ρ=0.74

585 K, ρ=0.61 450 K, ρ=0.74

Fig. 2.

CLaMS transport of CH

4

versus HALOE observations northward of 30

0

N between April, 10 and June 31 1997 at

θ= 450

and

θ = 585

K without mixing (upper panel) and with optimal mixing (lower panel).

for the onset of the final warming in the northern hemisphere (Manney and Sabutis, 2000).

The zonal distribution of the vortex remnants after the vor- tex breakup is mainly determined by the isentropic winds. In Fig. 3 the zonal mean winds at 450 and 585 K are shown for both considered periods. Owing to the long vortex lifetime in 1997, the vortex breakup at 585 K is accompanied by a transition from the winter to the summer circulation with no influence of the subtropical jet. On the other hand, the trans- port of vortex remnants at 450 K, is driven by westerly winds and influenced by the subtropical jet. The zonally averaged winds in 2000 show a similar structure although a strong sub- tropical jet is present even in the upper level and the summer circulation is still not fully developed owing to the earlier breakup time.

By performing the CLaMS simulations, we investigate now the meridional distribution of the vortex air and the in- fluence of mixing (i.e. of the irreversible part of transport) on a such distribution. CLaMS results for the zonally aver- aged meridional distribution of the vortex air after the vortex

breakup in 1997 are shown in Fig. 4. Results at 450 K (left) and 585 K (right) are plotted for tracer transport without mix- ing (top) and with an excessive mixing (bottom) correspond- ing to the mean distance between neighboring APs r

0

= 200 km. In the model, vortex air is defined as air masses bounded at the beginning of the simulation by the vortex edge identi- fied by the strongest PV gradient with respect to equivalent latitude (Nash et al., 1996). To mark vortex air, an additional, artificial tracer is used and initialized as 1 and 0 within and outside of the vortex, respectively. The subsequent transport (advection + mixing) of this tracer describes the zonal distri- bution of the vortex air.

The comparison between top and bottom panels of Fig. 4 shows a negligible influence of mixing on the zonal distri- bution of the vortex air. Even for CLaMS simulations with- out mixing, the zonal averaging of the (unmixed) air parcels leads to a similar meridional distribution of the vortex air as in the case where mixing was exaggerated. Thus, the large scale meridional transport of vortex air into the mid-latitudes is dominated by the chaotic advection (induced e.g. by plan- etary waves) rather than by mixing. The zonally averaged tracks of vortex remnants (black lines in Figs. 4 and 3) show at 585 K a stronger southward propagation than at 450 K. In agreement with the investigations of Piani et al. (2002), this propagation at 450 K is confined by the subtropical jet to lat- itudes poleward of 55

N. The southward transport of vortex air at 585 K is more effective and reaches about 40

N.

However, mixing may significantly influence the lifetime of vortex remnants, i.e., the time that is necessary to homoge- neously mix vortex air with ambient air. The temporally and spatially inhomogeneous CLaMS mixing is driven by inte- gral flow deformation that can be quantified in terms of the finite-time Lyapunov exponent λ . Thus, CLaMS produces high mixing intensity only in flow regions with sufficiently high values of λ. Fig. 5 shows the zonally averaged Lya- punov exponents λ calculated for each CLaMS AP over a time step ∆t = 12 hours. After the vortex breakup, the sum- mer circulation at 585 K is characterized by very low values of λ. Generally, the summer circulation in the middle strato- sphere can be understood as a “solid body rotation” with a negligible amount of local strain between the neighboring APs (Piani and Norton, 2002). Consequently, the regridding algorithm in CLaMS that is driven by the local deformation rates, indicates very weak mixing at this level that is in agree- ment with the diagnostic of atmospheric transport in terms of the effective diffusivity ite[]Haynes2000a.

To quantify the effect of mixing on the lifetime of vor- tex remnants the PDFs (probability density function) calcu- lated for CLaMS CH

4

distributions are shown in Fig. 6. The PDF is proportional to the area occupied by tracer values in a given range of mixing ratios (for details see e.g. Sparling, 2000). Here, the black lines denote the vortex edge trans- formed from the PV to CH

4

-field. Thus, the PDF of the CH

4

-values below these lines describe the contribution of vortex air to all air masses poleward of 30

N. The results

Atmos. Chem. Phys., 0000, 0001–12, 2003 www.atmos-chem-phys.org/0000/0001/

Fig. 2. CLaMS transport of CH4 versus HALOE observations northward of 30N between 10 April and 31 June 1997 atθ=450 andθ=585 K without mixing (upper panel) and with optimal mix- ing (lower panel).

3 Spatial distribution and lifetime of vortex remnants in 1997 and 2000

In 1997, the polar vortex was unusually long-lived (Coy et al., 1997). Even at the end of April, the vortex was very symmetrical around the pole and stable, and it did not break up until mid-May. On the other hand, the breakup of the vortex in 2000 was around mid-March, i.e. during a time pe- riod typical for the onset of the final warming in the northern hemisphere (Manney and Sabutis, 2000).

The zonal distribution of the vortex remnants after the vor- tex breakup is mainly determined by the isentropic winds. In Fig. 3 the zonal mean winds at 450 and 585 K are shown for both periods considered. Owing to the long vortex lifetime in 1997, the vortex breakup at 585 K is accompanied by a transition from the winter to the summer circulation with no

influence of the subtropical jet. On the other hand, the trans- port of vortex remnants at 450 K is driven by westerly winds and influenced by the subtropical jet. The zonally averaged winds in 2000 show a similar structure although a strong sub- tropical jet is present even in the upper level and the summer circulation is still not fully developed owing to the earlier breakup time.

By performing the CLaMS simulations, we now investi- gate the meridional distribution of the vortex air and the in- fluence of mixing (i.e. of the irreversible part of transport) on such a distribution. CLaMS results for the zonally aver- aged meridional distribution of the vortex air after the vortex breakup in 1997 are shown in Fig. 4. Results at 450 K (left) and 585 K (right) are plotted for tracer transport without mix- ing (top) and with excessive mixing (bottom) corresponding to the mean distance between neighboring APsr0=200 km.

In the model, vortex air is defined as air masses bounded at the beginning of the simulation by the vortex edge identified by the strongest PV gradient with respect to equivalent lati- tude (Nash et al., 1996). To mark vortex air, an additional, artificial tracer is used and initialized as 1 and 0 within and outside of the vortex, respectively. The subsequent transport (advection + mixing) of this tracer describes the zonal distri- bution of the vortex air.

The comparison between top and bottom panels of Fig. 4 shows a negligible influence of mixing on the zonal distri- bution of the vortex air. Even for CLaMS simulations with- out mixing, the zonal averaging of the (unmixed) air parcels leads to a similar meridional distribution of the vortex air as in the case where mixing was exaggerated. Thus, the large- scale meridional transport of vortex air into the mid-latitudes is dominated by the chaotic advection (induced e.g. by plan- etary waves) rather than by mixing. The zonally averaged tracks of vortex remnants (black lines in Figs. 4 and 3) show a stronger southward propagation at 585 K than at 450 K. In agreement with the investigations of Piani et al. (2002), this propagation at 450 K is confined by the subtropical jet to lat- itudes poleward of 55N. The southward transport of vortex air at 585 K is more effective and reaches about 40N.

However, mixing may significantly influence the lifetime of vortex remnants, i.e., the time that is necessary to homoge- neously mix vortex air with ambient air. The temporally and spatially inhomogeneous CLaMS mixing is driven by inte- gral flow deformation that can be quantified in terms of the finite-time Lyapunov exponentλ.

Thus, CLaMS produces high mixing intensity only in flow regions with sufficiently high values ofλ. Figure 5 shows the zonally averaged Lyapunov exponents λ calculated for each CLaMS AP over a time step 1t =12 hours. After the vortex breakup, the summer circulation at 585 K is char- acterized by very low values ofλ. Generally, the summer circulation in the middle stratosphere can be understood as a

“solid body rotation” with a negligible amount of local strain between the neighboring APs (Piani and Norton, 2002). Con- sequently, the regridding algorithm in CLaMS that is driven

Atmos. Chem. Phys., 3, 839–849, 2003 www.atmos-chem-phys.org/acp/3/839/

(5)

Konopka et al.: Dynamics and chemistry of vortex remnantsKonopka et al.: Dynamics and chemistry of vortex remnants 5 843 1997

20 30 40 50 60 70 80 90

Latitude [oN]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80 90

Latitude [oN]

Vortex breakup

Track of vortex remnants

20 30 40 50 60 70 80 90

Latitude [oN]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80 90

Latitude [oN]

01.05.97 01.06.97 01.07.97 time

Track of vortex remnants

Subtrop. jet

2000

20 30 40 50 60 70 80 90

Latitude [oN]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80 90

Latitude [oN]

Vortex breakup

Subtrop. jet Track of

vortex remnants

20 30 40 50 60 70 80 90

Latitude [oN]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80 90

Latitude [oN]

01.03.00 01.05.00

time

−20.0

−20.0

−15.0

−10.0

−5.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 U [m/s]

Vortex breakup

Subtrop. jet Track of

vortex remnants

Fig. 3. Zonal mean ECMWF winds atθ= 450(bottom panels) and 585 K (top panels) for the 1997 (left) and 2000 (right) period. The southward propagation of the vortex remnants is bounded by the subtropical jet. Their meridional tracks are derived from zonally averaged distribution of the vortex air (see Fig. 4).

show that the vortex remnants disappear significantly faster at 450 than at 585 K. The greater Lyapunov exponents in the lower level cause more stirring, greater local deforma- tion rates, and, consequently, more mixing. The lifetime of the vortex remnants at 450 and 585 K is of the order 5 and 10 weeks, respectively. The behavior of vortex remnants formed in March 2000 is similar but, due to an earlier vortex breakup, dominated until mid May by westerly winds, even at 585 K.

It should be emphasized that isentropic simulations may overestimate the lifetime of the vortex remnants, especially in the last phase of their existence when the remnants are los- ing their vertical coherence. Here, the vortex fragments may form elongated and slanted sheets of air, the so-called lam- inae (see e.g. Orsolini et al., 1995) with a very complicated contact surface separating the vortex from the mid latitude air (Haynes and Anglade, 1997). Thus, although the isentropic 2d simulations underestimate such contact surfaces and, con- sequently, overestimate the lifetime of the vortex remnants,

the ratio of their lifetimes is a rather reliable quantity. We conclude that the lifetime of the remnants at 585 K is by a factor of 2 longer than at 450 K.

4 Ozone chemistry in vortex remnants

In the previous section, we have shown that the vortex air trapped in the long-lived vortex remnants is well-isolated from the mid-latitudes. Here, we discuss some properties of the O3-chemistry observed and simulated in such remnants in spring/summer 1997 and 2000. It is a well-established fact that the polar O3 loss is due to halogen-catalyzed ozone loss with primary contributions of the reactive chlorine and bromine species which are activated heterogeneously on the surface of the polar stratospheric cloud particles, a primary component of which is nitric acid (HNO3) (Solomon, 1999).

Under Arctic conditions, the ozone loss terminates in spring, as chlorine is deactivated through the formation of ClONO2

www.atmos-chem-phys.org/0000/0001/ Atmos. Chem. Phys., 0000, 0001–12, 2003

Fig. 3. Zonal mean ECMWF winds atθ = 450 (bottom panels) and 585 K (top panels) for the 1997 (left) and 2000 (right) period. The southward propagation of the vortex remnants is bounded by the subtropical jet. Their meridional tracks are derived from zonally averaged distribution of the vortex air (see Fig. 4).

6 Konopka et al.: Dynamics and chemistry of vortex remnants

pure advection

mixing too large

20 30 40 50 60 70 80

Latitude [°N]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80

Latitude [°N]

Vortex breakup

Track of vortex remnants

Subtrop. jet

20 30 40 50 60 70 80

Latitude [°N]

20 30 40 50 60 70 80

Latitude [°N]

01.05.97 01.06.97 01.07.97 time

Track of vortex remnants

Subtrop. jet

20 30 40 50 60 70 80

Latitude [°N]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80

Latitude [°N]

Vortex breakup

Track of vortex remnants

20 30 40 50 60 70 80

Latitude [°N]

20 30 40 50 60 70 80

Latitude [°N]

01.05.97 01.06.97 01.07.97 time

100.

0.

20.

40.

60.

80.

100.

Part [%]

Track of vortex remnants

Fig. 4. Meridional (zonally averaged) contribution of vortex air atθ= 450(left) and 585 K (right) for CLaMS simulation without (top) and with excessive (bottom) mixing. The colors denote the zonally averaged percentage the vortex air changing from 100% (pure vortex air) to 0% (pure extra vortex air). The black lines denote the tracks of the vortex air masses during their transport into the mid latitudes (see also Fig. 3)

and a subsequent transformation to HCl (Douglass et al., 1995). The main source of NOx that is controlling the de- activation of ClOx is either the photolysis of HNO3, its re- action with OH, or its flux from mid-latitudes. During the period when ClONO2 is converted into HCl, the photolyti- cally or heterogenously induced reactivation of ClOx from ClONO2 is still possible.

Thus on the one hand, the formation of HCl in the vor- tex air determines when the chlorine-induced O3-destruction is terminated. On the other hand, the breakup of the vor- tex defines when chaotic advection and intense mixing dom- inate the transport of vortex air into the mid-latitudes. Fig. 7 shows the simulated HCl distributions (using the optimal mixing) two days after the vortex breakup in 1997 and 2000.

The black circles denote the corresponding HALOE obser- vations that in the vortex remnants agree fairly well with the CLaMS calculations. The high HCl mixing ratios that were observed in May 1997 indicate a completed chlorine deacti- vation whereas this process is still under way in March 2000.

As can be seen in Fig. 8, first at the end of April (i.e. 44

days after the vortex breakup) significantly enhenced mix- ing ratios of HCl become apparent in the simulated vortex remnants that were also observed by HALOE (M¨uller et al., 2002).

A more quantitative description of the deactivation process is given in Fig. 9 where the mean O3-loss rate (per day) av- eraged over the vortex air together with the partitioning of O3-loss into different chemical destruction cycles are shown.

To distinguish the vortex from the mid-latitude air, we use the transported (i.e. advected and mixed) PV fields. Then, the vortex air is defined as air masses with PV values larger than PV at the vortex edge (Nash et al., 1996) at the begin- ning of the simulation.

A comparison of the mean net O3-loss rate at 450 K dur- ing the time periods around the vortex breakup in 1997 and 2000 shows in March 2000 significantly faster O3-depletion than in May 1997. In particular, between the time of the vortex breakup around mid-March and mid-April 2000, the O3 destruction in the vortex remnants was still dominated by the halogen cycles (green line). By contrast, in mid-May

Atmos. Chem. Phys., 0000, 0001–12, 2003 www.atmos-chem-phys.org/0000/0001/

Fig. 4. Meridional (zonally averaged) contribution of vortex air atθ=450 (left) and 585 K (right) for CLaMS simulation without (top) and with excessive (bottom) mixing. The colors denote the zonally averaged percentage of the vortex air changing from 100% (pure vortex air) to 0% (pure extra vortex air). The black lines denote the tracks of the vortex air masses during their transport into the mid-latitudes (see also Fig. 3)

www.atmos-chem-phys.org/acp/3/839/ Atmos. Chem. Phys., 3, 839–849, 2003

(6)

844 Konopka et al.: Dynamics and chemistry of vortex remnants

Konopka et al.: Dynamics and chemistry of vortex remnants 7

20 30 40 50 60 70 80

Latitude [oN]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80

Latitude [oN]

Vortex breakup

20 30 40 50 60 70 80

Latitude [oN]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80

Latitude [oN]

01.05.97 01.06.97 01.07.97 time

1.0

0.0 0.1 0.3 0.4 0.6 0.7 0.9 1.0 mean λ [day−1]

Fig. 5. Zonally averaged Lyapunov exponent λ during the 1997 period calculated over a time step∆t =12 hours atθ = 450and 585 K. Note that the summer circulation at 585 K is characterized by very low values ofλ.

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 585 K θ = 585 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

Vortex edge

Vortex breakup

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 450 K θ = 450 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

01.05.97 01.06.97 01.07.97 time

−10.0

−10.0

−8.9

−7.7

−6.6

−5.4

−4.3

−3.1

−2.0 log(PDF)

Fig. 6. PDF of CH4 atθ = 450and 585 K calculated during the 1997 period. The black lines denote the mean CH4 at the vortex edge at the beginning of the simulation. The vortex edge was deter- mined by using the Nash criterion (Nash et al., 1996).

Fig. 7. HCl simulated with CLaMS at θ = 450 K 2 days after the vortex breakup at 12.05.1997 (top) and 17.03.2000 (bottom).

High values of HCl indicate a completed chlorine deactivation. The circles denote the HALOE observations

1997, these cycles are negligible, even shortly before the vor- tex breakup. As already mentioned, these differences can be explained by the fact that the chlorine deactivation due to for- mation of HCl is completed before the final vortex breakup in May 1997, whereas the formation of HCl in spring 2000 is finished in the vortex remnants in mid-April.

At 450 K the contribution of HO

x

(HO

x

=HO

2

+OH) out- weights the NO

x

-induced ozone destruction and is domi- nated by the direct reaction of HO

x

with ozone. Here, the ozone destruction driven by NO

x

that was formed by pho- tolytical decomposition of ClONO

2

(Toumi et al., 1993) is classified as NO

x

-induced ozone destruction and contributes up to 45% to the NO

x

-cycles around mid April (not shown in Fig. 9).

A similar analysis of the O

3

-chemistry in the long-lived vortex remnants at 585 K in 1997 shows that the most impor-

www.atmos-chem-phys.org/0000/0001/ Atmos. Chem. Phys., 0000, 0001–12, 2003

Fig. 5. Zonally averaged Lyapunov exponentλduring the 1997 period calculated over a time step1t =12 hours atθ =450 and 585 K. Note that the summer circulation at 585 K is characterized by very low values ofλ.

by the local deformation rates, indicates very weak mixing at this level that is in agreement with the diagnostic of atmo- spheric transport in terms of the effective diffusivity (Haynes and Shuckburgh, 1999).

To quantify the effect of mixing on the lifetime of vor- tex remnants the PDFs (probability density function) calcu- lated for CLaMS CH4 distributions are shown in Fig. 6. The PDF is proportional to the area occupied by tracer values in a given range of mixing ratios (for details see e.g. Sparling, 2000). Here, the black lines denote the vortex edge trans- formed from the PV to CH4 field. Thus, the PDFs of the CH4 values below these lines describe the contribution of vortex air to all air masses poleward of 30N. The results show that the vortex remnants disappear significantly faster at 450 than at 585 K. The greater Lyapunov exponents in the lower level cause more stirring, greater local deforma- tion rates, and, consequently, more mixing. The lifetime of the vortex remnants at 450 and 585 K is of the order 5 and 10 weeks, respectively. The behavior of vortex remnants formed in March 2000 is similar but, due to an earlier vortex breakup, dominated until mid-May by westerly winds, even at 585 K.

It should be emphasized that isentropic simulations may overestimate the lifetime of the vortex remnants, especially in the last phase of their existence when the remnants are los-

Konopka et al.: Dynamics and chemistry of vortex remnants 7

20 30 40 50 60 70 80

Latitude [oN]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80

Latitude [oN]

Vortex breakup

20 30 40 50 60 70 80

Latitude [oN]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80

Latitude [oN]

01.05.97 01.06.97 01.07.97 time

1.0

0.0 0.1 0.3 0.4 0.6 0.7 0.9 1.0 mean λ [day−1]

Fig. 5. Zonally averaged Lyapunov exponent λ during the 1997 period calculated over a time step∆t =12 hours atθ = 450and 585 K. Note that the summer circulation at 585 K is characterized by very low values ofλ.

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 585 K θ = 585 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

Vortex edge

Vortex breakup

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 450 K θ = 450 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

01.05.97 01.06.97 01.07.97 time

−10.0

−10.0

−8.9

−7.7

−6.6

−5.4

−4.3

−3.1

−2.0 log(PDF)

Fig. 6. PDF of CH4 atθ = 450 and 585 K calculated during the 1997 period. The black lines denote the mean CH4 at the vortex edge at the beginning of the simulation. The vortex edge was deter- mined by using the Nash criterion (Nash et al., 1996).

Fig. 7. HCl simulated with CLaMS atθ = 450 K 2 days after the vortex breakup at 12.05.1997 (top) and 17.03.2000 (bottom).

High values of HCl indicate a completed chlorine deactivation. The circles denote the HALOE observations

1997, these cycles are negligible, even shortly before the vor- tex breakup. As already mentioned, these differences can be explained by the fact that the chlorine deactivation due to for- mation of HCl is completed before the final vortex breakup in May 1997, whereas the formation of HCl in spring 2000 is finished in the vortex remnants in mid-April.

At 450 K the contribution of HO

x

(HO

x

=HO

2

+OH) out- weights the NO

x

-induced ozone destruction and is domi- nated by the direct reaction of HO

x

with ozone. Here, the ozone destruction driven by NO

x

that was formed by pho- tolytical decomposition of ClONO

2

(Toumi et al., 1993) is classified as NO

x

-induced ozone destruction and contributes up to 45% to the NO

x

-cycles around mid April (not shown in Fig. 9).

A similar analysis of the O

3

-chemistry in the long-lived vortex remnants at 585 K in 1997 shows that the most impor-

www.atmos-chem-phys.org/0000/0001/ Atmos. Chem. Phys., 0000, 0001–12, 2003

Fig. 6. PDF of CH4 atθ = 450 and 585 K calculated during the 1997 period. The black lines denote the mean CH4 at the vortex edge at the beginning of the simulation. The vortex edge was deter- mined by using the Nash criterion (Nash et al., 1996).

ing their vertical coherence. Here, the vortex fragments may form elongated and slanted sheets of air, the so-called lam- inae (see e.g. Orsolini et al., 1995) with a very complicated contact surface separating the vortex from the mid-latitude air (Haynes and Anglade, 1997). Thus, although the isen- tropic 2-D simulations underestimate such contact surfaces and, consequently, overestimate the lifetime of the vortex remnants, the ratio of their lifetimes is a rather reliable quan- tity. We conclude that the lifetime of the remnants at 585 K is longer than at 450 K by a factor of 2.

4 Ozone chemistry in vortex remnants

In the previous section, we showed that the vortex air trapped in the long-lived vortex remnants is well-isolated from the mid-latitudes. Here, we discuss some properties of the O3 chemistry observed and simulated in such remnants in spring/summer 1997 and 2000. It is a well-established fact that the polar O3 loss is due to halogen-catalyzed ozone loss with primary contributions of the reactive chlorine and bromine species which are activated heterogeneously on the surface of the polar stratospheric cloud particles, a primary component of which is nitric acid (HNO3) (Solomon, 1999).

Atmos. Chem. Phys., 3, 839–849, 2003 www.atmos-chem-phys.org/acp/3/839/

(7)

Konopka et al.: Dynamics and chemistry of vortex remnants 845

Konopka et al.: Dynamics and chemistry of vortex remnants 7

20 30 40 50 60 70 80

Latitude [oN]

θ = 585 K θ = 585 K

20 30 40 50 60 70 80

Latitude [oN]

Vortex breakup

20 30 40 50 60 70 80

Latitude [oN]

θ = 450 K θ = 450 K

20 30 40 50 60 70 80

Latitude [oN]

01.05.97 01.06.97 01.07.97 time

1.0

0.0 0.1 0.3 0.4 0.6 0.7 0.9 1.0 mean λ [day−1]

Fig. 5. Zonally averaged Lyapunov exponent λduring the 1997 period calculated over a time step∆t=12 hours atθ = 450and 585 K. Note that the summer circulation at 585 K is characterized by very low values ofλ.

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 585 K θ = 585 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

Vortex edge

Vortex breakup

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

θ = 450 K θ = 450 K

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

CH4 [ppmv]

01.05.97 01.06.97 01.07.97 time

−10.0

−10.0

−8.9

−7.7

−6.6

−5.4

−4.3

−3.1

−2.0 log(PDF)

Fig. 6. PDF of CH4 atθ = 450 and 585 K calculated during the 1997 period. The black lines denote the mean CH4 at the vortex edge at the beginning of the simulation. The vortex edge was deter- mined by using the Nash criterion (Nash et al., 1996).

Fig. 7. HCl simulated with CLaMS atθ = 450 K 2 days after the vortex breakup at 12.05.1997 (top) and 17.03.2000 (bottom).

High values of HCl indicate a completed chlorine deactivation. The circles denote the HALOE observations

1997, these cycles are negligible, even shortly before the vor- tex breakup. As already mentioned, these differences can be explained by the fact that the chlorine deactivation due to for- mation of HCl is completed before the final vortex breakup in May 1997, whereas the formation of HCl in spring 2000 is finished in the vortex remnants in mid-April.

At 450 K the contribution of HO

x

(HO

x

=HO

2

+OH) out- weights the NO

x

-induced ozone destruction and is domi- nated by the direct reaction of HO

x

with ozone. Here, the ozone destruction driven by NO

x

that was formed by pho- tolytical decomposition of ClONO

2

(Toumi et al., 1993) is classified as NO

x

-induced ozone destruction and contributes up to 45% to the NO

x

-cycles around mid April (not shown in Fig. 9).

A similar analysis of the O

3

-chemistry in the long-lived vortex remnants at 585 K in 1997 shows that the most impor-

www.atmos-chem-phys.org/0000/0001/ Atmos. Chem. Phys., 0000, 0001–12, 2003

Fig. 7. HCl simulated with CLaMS atθ = 450 K 2 days after the vortex breakup at 12.05.1997 (top) and 17.03.2000 (bottom).

High values of HCl indicate a completed chlorine deactivation. The circles denote the HALOE observations

Under Arctic conditions, the ozone loss terminates in spring, as chlorine is deactivated through the formation of ClONO2

and a subsequent transformation to HCl (Douglass et al., 1995). The main source of NOx which controls the deacti- vation of ClOx is either the photolysis of HNO3, its reaction with OH, or its flux from mid-latitudes. During the period when ClONO2 is converted into HCl, the photolytically or heterogenously induced reactivation of ClOx from ClONO2 is still possible.

Thus on the one hand, the formation of HCl in the vortex air determines when the chlorine-induced O3 destruction is terminated. On the other hand, the breakup of the vortex de- fines when chaotic advection and intense mixing dominate the transport of vortex air into the mid-latitudes. Figure 7 shows the simulated HCl distributions (using the optimal mixing) two days after the vortex breakup in 1997 and 2000.

8 Konopka et al.: Dynamics and chemistry of vortex remnants

Fig. 8. HCl simulated with CLaMS atθ = 450K on 29.04.2000, i.e. 44 days after the vortex breakup.

tant destruction cycle is the “summertime” NO

x

-chemistry (Hansen and Chipperfield, 1999; Fahey and Ravishankara, 1999). Furthermore, the fractional contribution of the dif- ferent cycles to the O

3

-loss in the vortex remnants does not significantly differ from the O

3

-loss partitioning calculated for extra-vortex air masses northward of 60

0

N (not shown).

So although the chemical composition of the coarsely-mixed vortex remnants still differs from the composition of the am- bient air, the contribution of the O

3

-loss cycles to the total O

3

-depletion is very similar in both types of air.

The fact that the deactivation process in spring 1997 was still completed in a well-isolated vortex indicates that at least in this year, the chlorine deactivation was decoupled from mixing and mainly driven by the in situ chemical produc- tion of NO

x

. On the other side, the incomplete deactivation shortly before the vortex breakup in March 2000 offers the possibility to study the influence of mixing on this process during the final vortex decay.

5 The impact of mixing

Using different values of the mixing parameters, we now study the influence of mixing on the accumulated ozone loss in the mid and high latitudes after the vortex breakup around mid March 2000 at the isentropic surface 450 K. The accu- mulated ozone loss is defined as the difference between the passively transported and chemically changed ozone. Here, this difference is determined for the time period between 10.02 and 31.05.2000 and averaged over all air masses pole- ward of 30

0

N. The results calculated for the optimal mixing are shown in Fig. 10 (red line).

In order to study the influence of halogen chemistry on the O

3

-loss, the black line describes the accumulated ozone loss calculated from a chemistry run without halogen cy- cles (Cl

y

=Br

y

=0). The yellow curve describes the accumu- lated ozone loss for a scenario with a strong denitrification as would be expected for Antarctic conditions (HNO

3≈2ppbv

in the vortex). The remaining curves describe the sensitivity of the reference simulation (red) on the intensity of mixing with effective diffusivity varying between 1.1 10

3

and 4.2 10

5

m

2

s

1

Thus, the halogen-induced mid-latitude ozone deficit in spring 2000, defined here as the difference between the red and black curves, can be divided into 3 phases (see dashed red and black lines in Fig. 10): Until mid-March ozone de- struction due to halogen chemistry occurs in a well-isolated Arctic vortex. From mid-March (vortex breakup) until mid- April the ozone-depleted air masses are transported into the mid-latitudes. By comparing the slopes of the red and black lines during this period, we conclude that the chlorine and bromine chemistry still destroy ozone in vortex remnants (see Fig. 9), albeit with a smaller intensity than in February and in early March. Here, the main source of active chlorine is the photolysis of ClONO

2

. Between mid-April and end of May, the slopes of the dashed black and red lines are compa- rable and, consequently, the halogen-induced ozone destruc- tion is negligible during this period. Thus, compared with the ozone distribution in a halogen-free stratosphere (black line), the mid-latitude ozone deficit at the isentropic level 450 K can be determined. Thus, the halogen-induced ozone deficit at the end of May amounts to about 12%. The contribution of the vortex remnants formed after the vortex breakup can be quantified to about 2%.

The sensitivity studies with respect to mixing show that the accumulated ozone loss does not change as long as the effec- tive diffusivity in the model is smaller than

3×104

m

2

/s. This value corresponds to a spatial resolution of the order 100 km.

Also, the deactivation of ClO

x

via formation of ClONO

2

is mainly due to photochemical decomposition of HNO

3

rather than due to chemistry induced by mixing of the activated vor- tex air with NO

x

-rich mid-latitude air. Only for the diffusion being greater than an unrealistically large critical value of

105

m

2

/s (corresponding to a spatial resolution of the order 200 km) mixing of mid-latitude air has a significant impact on ClONO

2

formation in vortex remnants.

Using an Eulerian model, Tan et al. (1998) postulated a much stronger influence of the numerical mixing on the chlorine deactivation. They concluded that Eulerian grid resolution better than 40 km is necessary to correctly de- scribe the deactivation process. Assuming that the numer- ical diffusivity is proportional to

r20/∆t, the discussed La-

grangian critical resolution

r0L =r0≈100

km can be trans- formed to an equivalent Eulerian resolution

r0E

by

r0E = r0L

p∆tE/∆tL

where

∆tL ≈ 24

hours and

∆tE ≈ 15

min are the typical Lagrangian and Eulerian time steps. Us- ing this crude estimate, we obtain

r0E ≈ 10

km that prob-

Atmos. Chem. Phys., 0000, 0001–12, 2003 www.atmos-chem-phys.org/0000/0001/

Fig. 8. HCl simulated with CLaMS atθ =450 K on 29.04.2000, i.e. 44 days after the vortex breakup.

The black circles denote the corresponding HALOE obser- vations that in the vortex remnants agree fairly well with the CLaMS calculations. The high HCl mixing ratios that were observed in May 1997 indicate a completed chlorine deacti- vation whereas this process is still under way in March 2000.

As can be seen in Fig. 8, first at the end of April (i.e. 44 days after the vortex breakup) significantly enhanced mix- ing ratios of HCl become apparent in the simulated vortex remnants that were also observed by HALOE (M¨uller et al., 2002).

A more quantitative description of the deactivation process is given in Fig. 9 where the mean O3 loss rate (per day) aver- aged over the vortex air together with the partitioning of O3 loss into different chemical destruction cycles are shown. To distinguish the vortex from the mid-latitude air, we use the transported (i.e. advected and mixed) PV fields. Then, the vortex air is defined as air masses with PV values larger than PV at the vortex edge (Nash et al., 1996) at the beginning of the simulation.

A comparison of the mean net O3 loss rate at 450 K dur- ing the time periods around the vortex breakup in 1997 and 2000 shows significantly faster O3 depletion in March 2000 than in May 1997. In particular, between the time of the vortex breakup around mid-March and mid-April 2000, the O3 destruction in the vortex remnants was still dominated by the halogen cycles (green line). By contrast, in mid-May 1997, these cycles are negligible, even shortly before the vor- tex breakup. As already mentioned, these differences can be explained by the fact that the chlorine deactivation due to for- mation of HCl is completed before the final vortex breakup in May 1997, whereas the formation of HCl in spring 2000 is finished in the vortex remnants in mid-April.

www.atmos-chem-phys.org/acp/3/839/ Atmos. Chem. Phys., 3, 839–849, 2003

Referenzen

ÄHNLICHE DOKUMENTE

The CLaMS accumulated ozone loss determined from the difference in relation to the passive ozone tracer O pass 3 is therefore a mixture of ozone depletion from air masses

These methods make it possible not only to obtain analytical stability criteria and to estimate limiting behavior (localization of self-excited and hidden attractors, study

Tropics: main flux into the stratosphere High latitudes: main flux from the stratosphere Weak vertical velocities (radiative equilibrium). adiabatic motion of air parcels

8a and b approximate the fraction of the CLaMS air parcels which was lifted from the lowest boundary layer to the middle and upper tropical troposphere using the deep

The scientific understanding of ozone loss in the Arctic and Antarctic stratosphere is built upon a combination of scientific discoveries and tested hypotheses extending

In the Northern Hemisphere (NH), the ILAS-II ozone data agree with the other data within ±10% (in terms of the absolute difference divided by its mean value) at altitudes between 11

During the descent of the aerosol layer, all stations ob- served only minor variations in backscatter ratio R of the parallel-polarised signal (Kiruna, Ny- ˚ Alesund) or unpo-

Fü Abbildung 6.6 wurde weiterhin der Einfluà der diabatischen Absinkraten untersucht (nicht abgebildet). Die Absinkraten wurden einmal mit dem Faktor 0,1 multipliziert,